Reliability of nutritional assessment in patients with gastrointestinal tumors

Patients with gastrointestinal cancer and malnutrition are less likely to tolerate major surgical procedures, radiotherapy or chemotherapy. In general, they display a higher incidence of complications such as infection, dehiscence and sepsis, which increases the length of stay and risk of death, and reduces quality of life. The aim of this review is to discuss the pros and cons of different points of view to assess nutritional risk in patients with gastrointestinal tract (GIT) tumors and their viability, considering the current understanding and screening approaches in the field. A better combination of anthropometric, laboratory and subjective evaluations is needed in patients with GIT cancer, since malnutrition in these patients is usually much more severe than in those patients with tumors at sites other than the GIT.


INTRODUCTION
C urrently, cancer is a major public health problem worldwide 1 .In addition, malnutrition and subsequent weight loss have long been among the leading causes of morbidity and mortality, as well as increased costs with other organs dysfunction associated to cancer patients undergoing surgery 2 .Malnutrition is defined as the energy, protein and other specific nutrients deficient state, which significantly modifies organic functions 3 .
Patients with gastrointestinal malignancy undergoing major elective procedures have a higher risk of postoperative complications and alterations resulting from their pre and post-admission nutritional status, particularly related to surgical stress, immune suppression induced by cancer or by blood transfusion.Among these factors, malnutrition is the most important due to its high prevalence and negative impact on clinical outcomes such as longer hospital stay 3 and mortality.The latter is much more related to malnutrition than cancer alone and can occur in 20% of cases 4 .Approximately half of the patients with malignancies has malnutrition, and in the case of gastrointestinal tract (GIT) tumors, the mortality rate varies from 30% to 50%, reaching 80% in cases of advanced pancreatic cancer 4,5 .
Several nutritional assessment methods can be used 5 , and must be sensitive enough to early identify changes according to specific nutritional imbalances.The method choice depends on the purpose of the assessment, prognosis or even on the response to nutritional interventions 2,4 .However, health professionals find it difficult to use most of the currently validated tools for nutritional assessment, due to limited time, method reproducibility, organization or cost [6][7][8] .Thus, all currently considered parameters show some sort of limitation to accurately assess the state nutritional 6 .In the absence of a gold standard, the option for the assessment tool and nutritional classification will depend on the institution and the target population in question, as well as on the resources available [8][9][10] .Although the use of indices and multivariate scores is often regarded as the solution to the lack of standardized and reliable evaluation, this is only a possibility 10 .DOI: 10.1590/0100-69912016003006
Therefore, in the daily practice of oncology, the definition of an appreciable and simple to apply nutritional assessment tool is necessary to identify nutritional risk patients and thus determine the best approach and appropriate nutritional support 8 .The objective of this review is to present an overview of the methods and tools used to determine nutritional risk, considering the pros and cons when applied to patients with GIT cancer.

METHODS
We systematically identified studies on nutritional status of patients with GIT cancer through the PubMed and MEDLINE databases.We researched articles published in the last ten years by combining the terms "nutritional assessment", "GI cancer", "gastrointestinal tract", "gastric cancer", "oesophageal cancer" and "pancreatic cancer".We considered for evaluation only complete articles with those terms in English or Portuguese.We identified additional articles from citations in the articles evaluated.

General review of nutritional assessment
It was in the 1950s that authors first published research related to nutritional assessment procedures.
Between 1960 and 1980, malnutrition markers have emerged to evaluate surgical patients and new concepts and nutritional assessment methods have been developed 11 .In the following decades, the researchers analyzed the relevance of functional indices and combinations of clinical and laboratory parameters existing in an attempt to better predict nutritional risk.A new concept of body composition was defined from the use of new and more complex equipment and methods of assessment, but still considering subjective concepts 7 .From the beginning of this century, attempts have been made to demonstrate the nutritional assessment method that would be more accurate and reliable for certain types of patients or specific clinical conditions 3,8 .

Subjective methods
In 1980, Detsky et al. described the Subjective Global Assessment (SGA) used to assess preoperative patients with GIT tumors (n = 202) undergoing major surgical procedures; They have shown that SGA could be easily applied and considered it a valid and reliable method to estimate the surgical patients' nutritional status 12 .
Other authors have published several articles supporting the SGA in determining nutritional status, which differed from other methods in considering not only changes in body composition, but also functional changes.In addition, SGA is a simple, inexpensive, non-invasive method and can be performed at bedside.Correct guidance on the SGA application is essential, since its accuracy depends on the observer's ability to detect subjectively significant nutritional changes 8,13,14 .
Subsequently, the SGA has undergone modifications and adaptations developed specifically to meet the oncological patient characteristics.Questions about symptoms of nutritional impact and resulting from the tumor itself or from the imposed treatment 8 were included and answered by the patient, becoming known as the Patient-generated SGA (PG-SGA).The main introduced difference was a numerical score that allows to better identify patients at nutritional risk and estimate the time required for re-evaluation.
In some multi-center studies on nutritional assessment of hospitalized patients using SGA, different results have been reported, particularly in patients with GIT cancer.Poziomyck et al. found 66% of malnourished patients in surgical cases of upper GIT tumors 8 , while Bragagnolo et al. showed 77% of malnourished patients in a similar sample 15 .
In another study involving 80 patients with GIT tumors, mainly colorectal, Cid Conde et al. found 50% of malnutrition by SGA 16 , a result that were higher (70%) in another study with a similar sample 17 .Wu et al. had higher incidence of complications and longer hospital stay the worse the level of SGA in patients undergoing major procedures for GIT cancer (mainly gastric) 18 .These results were also supported by our series of patients with esophageal, stomach and pan-creas tumors 8 .Moreover, in patients with esophageal or stomach cancer, SGA appears to be associated with the Glasgow Prognostic score (GPS) 19 .A study comparing PG-SGA with the Mini Nutritional Assessment (MNA) revealed that these tools seem appropriate to define elderly patients as malnourished 10 .GPS and MNA concepts are described in Score Methods later in the text.

Anthropometric methods
The accuracy and reproducibility of anthropometric measurements may be affected by the equipment calibration, examiner and parameters used for the predictive equations 20 .Several essentially objective nutritional assessment tools have been used in clinical practice, each with its own characteristics 13 .Body weight is as simple and commonly used measure in clinical practice.However, it does not discriminate mass from fat, muscle, bone or extracellular fluids.
Thus, it must be used with caution, since sharp alterations may reflect changes in hydration status, and not necessarily change in cell mass 21 .

Renehan et al. demonstrated that increased
body mass index (BMI) on the order of 5 kg/m 2 in both genders was strongly associated with esophageal adenocarcinoma 22 .Excess weight, visceral fat and abdominal obesity appears to be more disturbing than subcutaneous fat, and any further increase in BMI confers increased risk of developing colorectal cancer 19 , which, however, has not been confirmed in other studies with this tool 8,15 .
Functional markers are of particular importance, since they correlate well with clinical complications 20 .They may be more sensitive and relevant indicators of changes in nutritional state or response to additional support in the short term than conventional methods 8 .Loss of muscle function is indicative of malnutrition, particularly the loss of lean body mass.Usually expressed by the handgrip strength, it is important to determine the function and the ability of skeletal muscle.The authors consider this as evidence of compromised nutritional status as responsible for the loss of skeletal muscle function and, consequently, loss of   8,15,20 .Recently, the measurement of the thickness of the adductor pollicis muscle (APM) was standardized to anthropometric parameters relating to age, gender and physique 8 .APM has been used to indirectly determine the nutritional status 14,20,23 , being considered as one of the best single predictors of mortality in a recent study with patients undergoing resection of upper GIT tumor 8 .
The bioimpedance analysis (BIA) uses the measured phase angle, which is the result of the electric current stored in cell membranes.However, more accurate results depend on regression equations and lower values indicate reduction in cell integrity or cellular death 24 .
Some authors also use weight loss as a nutritional screening marker.In a study of patients with esophageal cancer, Van der Schaaf et al. found that preoperative loss weight exceeding 10% was associated with a reduction of the overall five-year survival after resection, but not with increased risk of postoperative complications 25 .

Laboratory Methods
Albumin and other proteins used as nutritional markers can be affected by many factors and clinical conditions such as inflammation, malnutrition, diabetes, liver disease or surgical trauma.But they also have been used to assess overall nutritional status, severity, progression and prognosis of the disease 26 , assuming that plasma levels indeed reflect the rate of synthesis 27-29.However, other factors such as liver function, inflammation markers and endocrine stress result in increased levels of cortisol, which also affects albumin regulation 27 .
Serum albumin has also been described as an independent survival prognostic factor in many tumors, displaying an inverse relation to complications and length of postoperative hospitalization or intensive care, mortality, and resumption of oral intake 26,27 .
Decreased serum albumin also proved to be an independent prognostic factor for cancer patients with unknown primary site 26 , but further clinical trials are needed to better define the baseline risks in patients with cancer 26-29.Recently, a significant association between increased C-reactive protein (CRP) and poor clinical outcome has been demonstrated in patients with pancreatic cancer 30 .CRP has also been shown to be an independent prognostic indicator in colorectal carcinoma 31 .

Score Methods
A number of studies have consistently shown that no method or tool alone is enough to predict nutritional status 3 .The mini nutritional assessment (MNA) classified as normal, borderline or malnutrition in the elderly involves anthropometric measurements, overall evaluation, dietary questionnaire and subjective evaluation 32 .
A cross-sectional study with elderly patients (n = 109) observed that combined arm circumference (AC) and BMI allowed to predict the MNA classification 33 .In another study evaluating elderly patients with hepatocellular carcinoma, the results suggested that MNA was adequate to identify the risk of deterioration in the quality of life and functional status, and to determine the risk of malnutrition 23 .
The Nutritional Risk Index (NRI) is calculated by the equation of serum albumin and weight ratio 3,34 ; the levels of serum protein and albumin significantly correlated with malnutrition, but not with subgroups of SGA or Nutritional Risk Screening 2002 (NRS-2002) 34 .
The Glasgow Prognostic Score (GPS) has been used to determine long-term outcome (survival) in cases of curable gastric cancer 34 , according to the degree of inflammation inferred by the CRP and albumin levels,with scores ranging from from 0 to 2 35 .
It may be useful in determining the nutritional status, since inflammation is a relevant factor in the development of cachexia, though not yet evaluated in the short term 34 .
In a study with 74 patients, 54 (72%) with GIT tumors, the Nutrition Inflammatory Index (NII) was an alternative method for biochemical nutritional assessment and monitoring of patients with cancer and systemic inflammation 36,37 .
The NRS-2002 is a nutritional and disease severity score, being the preferred method for evaluating patients at risk or malnourished and for selecting those that could benefit from nutritional support 3 .This nutritional screening was directly related to tumor stage in 100 newly diagnosed patients with stomach cancer, and inversely correlated with quality of life, making it a useful tool to identify patients in need of nutritional support throughout treatment 38 .

Poziomyck
Reliability of nutritional assessment in patients with gastrointestinal tumors  to better predict postoperative morbidity and mortality in patients with gastrointestinal cancer.
In summary, the GPS score, the PG-SGA and some anthropometric parameters are considered suitable for chronic and cancer patients in general.However, a better combination of laboratory, anthropometric and subjective evaluations is required, considering an instrument more focused in GIT cancer patients, since malnutrition in these patients is much more severe compared with the one in patients with tumors in other locations. 189-197

Table 1 .
Main objective and subjective methods used for nutritional assessment.

Table 2 .
Advantages and disadvantages of subjective methods and nutritional assessment.

Table 3 .
Advantages and disadvantages of anthropometric methods.

Table 4 .
Advantages and disadvantages of laboratory methods.
proaches are necessary to better assess the nutritional status in patients with GIT tumors, especially considering those patients with upper GIT tumors, who are more malnourished, more immunocompromised and at increased risk of morbidity and mortality, as recently demonstrated in our series 8 .To date, as far as we know, there is insufficient data to establish a consensus for this group of patients.Therefore, it would be interesting to simulate, add or combine features already validated with objective variables to test a single questionnaire specifically designed189-197

Table 5 .
Advantages and disadvantages of nutritional scores.
*Developed by SSORG (Southern Surgical Oncology Research Group)